views:

68

answers:

1

I'm trying to use rrdtool to monitor Access Points and what I'd like is to have separate rrd file for each access point, which is something I'm not sure how to do. Anyway if I can do that then for each site I'd be able to get a graph from different rrd databases according to site location. However when I want to see a company level graph I'd like to aggregate data across multiple rrd databases and get that to show on one graph, so if bandwidth is measured for two devices in two separate rrd databases then I would like to get an "average" of these two data sources and show it in my graph for the site that has these access points. Is this possible? I'm quite new to thinking in RRD way and rrdtool so please do let me know if there are better ways of doing this.

Also how RRD uses space internally? From what I read so far, there are people saying the size of file never gets bigger for RRD database. On the other side people asking about how much of file size it would accumulate over years. So I'm kind of confused here. I thought it would be holding stuff in memory and writing to disk based on consolidated functions.

Can I generate pie charts from rrdtool as well? I need to find number of users connected to a access point and it would be good if I can show that as a pie chart for total number of users connected to an access point at any given time for a given site. For instance,

access point 1: 20 access poin 2: 40 access poin 3: 1

If I can generate a pie chart for that it would be sliced according to the number of users.

Sorry it's quite a few questions. If rrdtool doesn't make a big difference then I might as well use Mysql as I have running mysql server in production. And I can produce graphs on the fly using some funky flash stuff too. If someone can enlighten me on pros and cons of using RRD over any RDBMS for time series data that would be amazing.

Many Thanks guys!!

A: 
  1. You can aggregate data from multiple RRDs into one graph; you'd use the CDEF command in your rrdgraph statement to combine DEFs from individual databases.

  2. rrd files stay the same size unless you explicitly resize them by adding rows. Older data is aged out and replaced with new data. (Hence the name "round robin database".)

  3. pie charts...I dunno. :) I've never seen it, but that certainly doesn't mean it's not possible.

Have you read the basic tutorial? http://oss.oetiker.ch/rrdtool/tut/rrdtutorial.en.html That might help you decide what to do.

gorthx
Yes I did look into some of the tutorials mate, thanks for your responses as well. Still I don't get how you define older data, in case if I'm saving data every 5 mins to monitor the heartbeat of an access point from now, and if im archiving an average for hourly, weekly, monthly and yearly basis, then when archives are created in the actual disk they would consume extra space isn't it.
pravin
The archives are defined when you create the original rrd file. You configure each RRA ('round robin archive') within an rrd file separately, defining (following your example) how many 5-minute averages you save, how many hourly averages you save, how many weekly, etc etc. That's all laid out when you create the file, so the file stays the same size.
gorthx
Alright that makes sense. I've been fiddling with it today and kind of figured out but still i have problem getting my head around "step:rows" when defining RRA. I guess more the step is then resolution increases for that archive. Is that right? Cheers again. appreciate your responses!
pravin